Local Patterns of Gradients for Face Recognition - AGPIG Accéder directement au contenu
Article Dans Une Revue IEEE Transactions on Information Forensics and Security Année : 2015

Local Patterns of Gradients for Face Recognition

Résumé

We present a novel feature extraction method namedas local patterns of gradients (LPOG) for robust face recognition.LPOG uses block-wised elliptical local binary patterns (BELBP),a refined variant of ELBP, and local phase quantization (LPQ)operators directly on gradient images for capturing local texturepatterns to build up a feature vector of a face image. Fromone input image, two directional gradient images are computed.A symmetric pair of BELBP and a LPQ operator are thenseparately applied upon each gradient image to generate localpatterns images. Histogram sequences of local patterns images’non-overlapped sub-regions are finally concatenated to form theLPOG vector for the given image. Based on LPOG descriptor, wepropose a novel face recognition system which exploits WhitenedPrincipal Component Analysis (WPCA) for dimension reductionand weighted angle based distance for classification. Experimentalresults on three large public databases (FERET, AR andSCface) prove that LPOG WPCA system is robust against a widerange of challenges such as: illumination, expression, occlusion,pose, time-lapse variations and low resolution. In addition, comparisonwith other systems shows that LPOG WPCA significantlyoutperforms state-of-the-art methods. Computationally, timingbenchmarks also demonstrate that our LPOG method is fasterthan many advanced feature extraction algorithms and can beapplied in real-world applications.
Fichier non déposé

Dates et versions

hal-01187416 , version 1 (26-08-2015)

Identifiants

Citer

Huu Tuan Nguyen, Alice Caplier. Local Patterns of Gradients for Face Recognition. IEEE Transactions on Information Forensics and Security, 2015, 10 (8), pp.1739-1751. ⟨10.1109/TIFS.2015.2426144⟩. ⟨hal-01187416⟩
289 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More